Skip to content

Commit

Permalink
added leptonica support and additional interfaces
Browse files Browse the repository at this point in the history 
git-svn-id: https://tesseract-ocr.googlecode.com/svn/trunk@135 d0cd1f9f-072b-0410-8dd7-cf729c803f20
  • Loading branch information
theraysmith committed Feb 1, 2008
1 parent e8787f3 commit d543e8c
Show file tree
Hide file tree
Showing 2 changed files with 134 additions and 21 deletions.
105 changes: 93 additions & 12 deletions ccmain/baseapi.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -19,6 +19,22 @@

#include "baseapi.h"


// Include automatically generated configuration file if running autoconf.
#ifdef HAVE_CONFIG_H
#include "config_auto.h"
#endif

#ifdef HAVE_LIBLEPT
// The jpeg library still has INT32 as long, which is no good for 64 bit.
#define INT32 WRONGINT32
#define BOX LEPT_BOX
// Include leptonica library only if autoconf (or makefile etc) tell us to.
#include "allheaders.h"
#undef BOX
#undef INT32
#endif

#include "tessedit.h"
#include "ocrclass.h"
#include "pageres.h"
Expand All @@ -27,20 +43,19 @@
#include "applybox.h"
#include "pgedit.h"
#include "varabled.h"
#include "variables.h"
#include "output.h"
#include "globals.h"
#include "adaptmatch.h"
#include "edgblob.h"
#include "tessbox.h"
#include "tordvars.h"
#include "tessvars.h"
#include "imgs.h"
#include "makerow.h"
#include "output.h"
#include "tstruct.h"
#include "tessout.h"
#include "tface.h"
#include "adaptmatch.h"
#include "permute.h"

BOOL_VAR(tessedit_resegment_from_boxes, FALSE,
"Take segmentation and labeling from box file");
Expand All @@ -52,6 +67,22 @@ const int kMinRectSize = 10;

static STRING input_file = "noname.tif";

// Set the value of an internal "variable" (of either old or new types).
// Supply the name of the variable and the value as a string, just as
// you would in a config file.
// Returns false if the name lookup failed.
bool TessBaseAPI::SetVariable(const char* variable, const char* value) {
if (set_new_style_variable(variable, value))
return true;
return set_old_style_variable(variable, value);
}

void TessBaseAPI::SimpleInit(const char* datapath,
const char* language,
bool numeric_mode) {
InitWithLanguage(datapath, NULL, language, NULL, numeric_mode, 0, NULL);
}

// Start tesseract.
// The datapath must be the name of the data directory or some other file
// in which the data directory resides (for instance argv[0].)
Expand All @@ -75,7 +106,7 @@ int TessBaseAPI::Init(const char* datapath, const char* outputbase,
// Start tesseract.
// Similar to Init() except that it is possible to specify the language.
// Language is the code of the language for which the data will be loaded.
// (Codes follow ISO 639-2.) If it is NULL, english (eng) will be loaded.
// (Codes follow ISO 639-3.) If it is NULL, english (eng) will be loaded.
int TessBaseAPI::InitWithLanguage(const char* datapath, const char* outputbase,
const char* language, const char* configfile,
bool numeric_mode, int argc, char* argv[]) {
Expand Down Expand Up @@ -192,6 +223,14 @@ void TessBaseAPI::DumpPGM(const char* filename) {
fclose(fp);
}

#ifdef HAVE_LIBLEPT
// ONLY available if you have Leptonica installed.
// Get a copy of the thresholded global image from Tesseract.
Pix* TessBaseAPI::GetTesseractImage() {
return page_image.ToPix();
}
#endif // HAVE_LIBLEPT

// Copy the given image rectangle to Tesseract, with adaptive thresholding
// if the image is not already binary.
void TessBaseAPI::CopyImageToTesseract(const unsigned char* imagedata,
Expand Down Expand Up @@ -250,6 +289,10 @@ void TessBaseAPI::OtsuThreshold(const unsigned char* imagedata,
int H;
int best_omega_0;
int best_t = OtsuStats(histogram, &H, &best_omega_0);
if (best_omega_0 == 0 || best_omega_0 == H) {
// This channel is empty.
continue;
}
// To be a convincing foreground we must have a small fraction of H
// or to be a convincing background we must have a large fraction of H.
// In between we assume this channel contains no thresholding information.
Expand Down Expand Up @@ -293,14 +336,14 @@ void TessBaseAPI::HistogramRect(const unsigned char* imagedata,
int* histogram) {
int width = right - left;
memset(histogram, 0, sizeof(*histogram) * 256);
const unsigned char* pix = imagedata +
top*bytes_per_line +
left*bytes_per_pixel;
const unsigned char* pixels = imagedata +
top*bytes_per_line +
left*bytes_per_pixel;
for (int y = top; y < bottom; ++y) {
for (int x = 0; x < width; ++x) {
++histogram[pix[x * bytes_per_pixel]];
++histogram[pixels[x * bytes_per_pixel]];
}
pix += bytes_per_line;
pixels += bytes_per_line;
}
}

Expand Down Expand Up @@ -420,8 +463,6 @@ void TessBaseAPI::FindLines(BLOCK_LIST* block_list) {
PAGE_RES* TessBaseAPI::Recognize(BLOCK_LIST* block_list, ETEXT_DESC* monitor) {
if (tessedit_resegment_from_boxes)
apply_boxes(block_list);
if (edit_variables)
start_variables_editor();

PAGE_RES* page_res = new PAGE_RES(block_list);
if (interactive_mode) {
Expand Down Expand Up @@ -455,6 +496,41 @@ int TessBaseAPI::TextLength(PAGE_RES* page_res) {
return total_length;
}

// Returns an array of all word confidences, terminated by -1.
int* TessBaseAPI::AllTextConfidences(PAGE_RES* page_res) {
if (!page_res) return NULL;
int n_word = 0;
PAGE_RES_IT res_it(page_res);
for (res_it.restart_page(); res_it.word () != NULL; res_it.forward())
n_word++;

int* conf = new int[n_word+1];
n_word = 0;
for (res_it.restart_page(); res_it.word () != NULL; res_it.forward()) {
WERD_RES *word = res_it.word();
WERD_CHOICE* choice = word->best_choice;
int w_conf = static_cast<int>(100 + 5 * choice->certainty());
// This is the eq for converting Tesseract confidence to 1..100
if (w_conf < 0) w_conf = 0;
if (w_conf > 100) w_conf = 100;
conf[n_word++] = w_conf;
}
conf[n_word] = -1;
return conf;
}

// Returns the average word confidence for Tesseract page result.
int TessBaseAPI::TextConf(PAGE_RES* page_res) {
int* conf = AllTextConfidences(page_res);
if (!conf) return 0;
int sum = 0;
int *pt = conf;
while (*pt >= 0) sum += *pt++;
if (pt != conf) sum /= pt - conf;
delete [] conf;
return sum;
}

// Make a text string from the internal data structures.
// The input page_res is deleted.
char* TessBaseAPI::TesseractToText(PAGE_RES* page_res) {
Expand Down Expand Up @@ -686,7 +762,6 @@ char* TessBaseAPI::TesseractToUNLV(PAGE_RES* page_res) {
}
return NULL;
}

// ____________________________________________________________________________
// Ocropus add-ons.

Expand Down Expand Up @@ -1009,3 +1084,9 @@ int TessBaseAPI::TesseractExtractResult(char** string,
}
return n;
}

// Check whether a word is valid according to Tesseract's language model
// returns 0 if the string is invalid, non-zero if valid
int TessBaseAPI::IsValidWord(const char *string) {
return valid_word(string);
}
50 changes: 41 additions & 9 deletions ccmain/baseapi.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -22,14 +22,31 @@

class PAGE_RES;
class BLOCK_LIST;
class IMAGE;
struct Pix;

// Base class for all tesseract APIs.
// Specific classes can add ability to work on different inputs or produce
// different outputs.

class TessBaseAPI {
public:
// Set the value of an internal "variable" (of either old or new types).
// Supply the name of the variable and the value as a string, just as
// you would in a config file.
// Returns false if the name lookup failed.
// For most variables, it is wise to set them before calling Init.
// Eg TessBaseAPI::SetVariable("tessedit_char_blacklist", "xyz");
static bool SetVariable(const char* variable, const char* value);

// Start tesseract.
// TODO(???): Make tesseract thread-safe, and then the init functions will
// return an instance of tesseract, and most of the other methods will become
// regular methods.
static void SimpleInit(const char* datapath, // Path to tessdata-no ending /.
const char* language, // ISO 639-3 string or NULL.
bool numeric_mode);

// The datapath must be the name of the data directory or some other file
// in which the data directory resides (for instance argv[0].)
// The configfile is the name of a file in the tessconfigs directory
Expand All @@ -49,7 +66,7 @@ class TessBaseAPI {
// Start tesseract.
// Similar to Init() except that it is possible to specify the language.
// Language is the code of the language for which the data will be loaded.
// (Codes follow ISO 639-2.) If it is NULL, english (eng) will be loaded.
// (Codes follow ISO 639-3.) If it is NULL, english (eng) will be loaded.
static int InitWithLanguage(const char* datapath, const char* outputbase,
const char* language, const char* configfile,
bool numeric_mode, int argc, char* argv[]);
Expand Down Expand Up @@ -97,6 +114,22 @@ class TessBaseAPI {
// Dump the internal binary image to a PGM file.
static void DumpPGM(const char* filename);

// Get a copy of the thresholded global image from Tesseract.
// Caller takes ownership of the Pix and must pixDestroy it.
// May be called before or after RecognizeText, or after TesseractRect.
static Pix* GetTesseractImage();

// Compute the Otsu threshold(s) for the given histogram.
// Also returns H = total count in histogram, and
// omega0 = count of histogram below threshold.
static int OtsuStats(const int* histogram,
int* H_out,
int* omega0_out);

// Check whether a word is valid according to Tesseract's language model
// returns 0 if the string is invalid, non-zero if valid
static int IsValidWord(const char *string);

protected:
// Copy the given image rectangle to Tesseract, with adaptive thresholding
// if the image is not already binary.
Expand Down Expand Up @@ -132,13 +165,6 @@ class TessBaseAPI {
int left, int top, int right, int bottom,
int* histogram);

// Compute the Otsu threshold(s) for the given histogram.
// Also returns H = total count in histogram, and
// omega0 = count of histogram below threshold.
static int OtsuStats(const int* histogram,
int* H_out,
int* omega0_out);

// Threshold the given grey or color image into the tesseract global
// image ready for recognition. Requires thresholds and hi_value
// produced by OtsuThreshold above.
Expand Down Expand Up @@ -170,6 +196,12 @@ class TessBaseAPI {

// Return the maximum length that the output text string might occupy.
static int TextLength(PAGE_RES* page_res);
// Returns the (average) confidence value between 0 and 100.
// The input page_res is NOT deleted.
static int TextConf(PAGE_RES* page_res);
// Returns all word confidences (between 0 and 100) in an array, terminated
// by -1. The calling function must delete [] after use.
static int* AllTextConfidences(PAGE_RES* page_res);
// Convert (and free) the internal data structures into a text string.
static char* TesseractToText(PAGE_RES* page_res);
// Make a text string from the internal data structures.
Expand Down Expand Up @@ -226,7 +258,7 @@ class TessBaseAPI {
// The segdata values are RGB triples, with distinct R values denoting distinct
// "blocks" or "columns" and distinct GB pairs denoting distinct lines.
// Lexicographic ordering of the RGB triples corresponds to text output order, with
// linebreaks inserted between distinct GB values and double blank lines between
// linebreaks inserted between distinct GB values and double blank lines between
// distinct R values.
static int SetPageSegmentation(const unsigned char *segdata,
int bytes_per_pixel, /* must be 3 */
Expand Down

0 comments on commit d543e8c

Please sign in to comment.